Epistemic Separability and Everettian Branches: A Critique of Sebens and Carroll